Image encoder, image decoder, image encoding method, and image decoding method

Abstract

An input image signal 101 is divided into MC block units and, when coding processing is performed in these divided units, a motion compensation section 107 generates a motion-compensated prediction image 106 a by detecting movement amounts in predetermined MC block units, a smoothing filter section 124 performs, with respect to the prediction image 106 a , smoothing of pixels located at the boundaries of adjoining MC blocks on the basis of predetermined evaluation criteria, and a prediction residual signal 108 , which is obtained from the difference between the prediction image 106 b obtained by the smoothing, and the input image (input image signal 101 ), is encoded. Accordingly, it is possible to use relatively straightforward computation to perform processing, with respect to a prediction frame image generated by block-unit motion-compensated interframe prediction (MC), to adaptively smooth a discontinuous waveform generated between MC blocks of the prediction frame image, whereby the efficiency of low bit rate coding that employs interframe MC is improved.

1. An image coding apparatus comprising:
 motion compensation predicting means for generating a motion-compensated prediction image by detecting movement amounts in predetermined first partial image regions of an input image with respect to a reference image; 
 smoothing means for performing smoothing of pixels located at the boundaries of adjoining images of the first partial image regions, with respect to the motion-compensated prediction image obtained by the motion compensation predicting means; and 
 prediction residual coding means for coding a prediction residual signal obtained from a difference between the input image and the smoothed prediction image, 
 wherein the smoothing means further comprises,
 activity level setting means for determining an activity level of second partial image regions of the input image; 
 adaptive smoothing means for establishing an intensity of the smoothing on a basis of the activity level set with said activity level setting means, and for performing smoothing of pixels located at the boundaries between adjoining images of the first partial image regions; and 
 post-processing means for performing suppression processing of a result of smoothing by using a threshold value established in accordance with said activity level, 
 
 wherein the first partial image regions are larger in size than the second partial image regions, so that at least one second partial image region fits into one of the first partial image regions, and at least one of the second partial image regions overlap a boundary of two adjacent first partial image regions when determining said activity level by the activity level setting means, and the second partial image regions are arranged in a cross-like pattern, having a central block with two blocks arranged above and below the central block, and two other blocks arranged to the left and the right of the central block. 
 
     
     
       2. The image coding apparatus as defined in  claim 1  wherein the motion compensation predicting means detects the movement amount by using encoded local decoding image data as the reference image. 
     
     
       3. The image coding apparatus as defined in  claim 1  wherein the smoothing means performs the smoothing in accordance with a difference value for the movement amount between the adjoining images of the first partial image regions. 
     
     
       4. The image coding apparatus as defined in  claim 1  wherein the smoothing means performs the smoothing in accordance with a difference in a coding method between the adjoining images of the first partial image regions. 
     
     
       5. The image coding apparatus as defined in  claim 1  wherein the smoothing means performs the smoothing based on whether the adjoining images of the first partial image regions have been predicted by bidirectional prediction, by unidirectional prediction, or by I-frame prediction. 
     
     
       6. The image coding apparatus as defined in  claim 1  wherein the smoothing means performs the smoothing in accordance with an error margin amount between the prediction image prior to smoothing, and the reference image from which the prediction image obtained by the motion compensation predicting means is generated. 
     
     
       7. The image coding apparatus as defined in  claim 1 , wherein said activity level of the second partial image regions is determined on a basis of at least one of: a difference value for the movement amount between the adjoining images of the first partial image regions, a difference in coding method between the adjoining images of the first partial image regions, a difference in image prediction method between the adjoining images of the first partial image regions, and an error margin amount between the prediction image prior to smoothing, and the reference image from which the prediction image obtained by the motion compensation predicting means is generated. 
     
     
       8. The image coding apparatus as defined in  claim 7  wherein the activity level setting means sets the activity level for individual second partial image regions and inputs to the adaptive smoothing means the larger value of values for the activity level obtained by comparing the first partial image regions which are subjected to smoothing. 
     
     
       9. The image coding apparatus as defined in  claim 7  wherein the activity level setting means sets the activity level for boundaries between the first partial image regions and inputs the activity level to the adaptive smoothing means. 
     
     
       10. The image coding apparatus as defined in  claim 7  wherein the adaptive smoothing means changes the number of pixels subjected to smoothing in accordance with the activity level obtained by the activity level setting means. 
     
     
       11. The image coding apparatus as defined in  claim 7  wherein the adaptive smoothing means switches filter characteristics for performing the smoothing in accordance with the activity level obtained by the activity level setting means. 
     
     
       12. The image coding apparatus as defined in  claim 11  wherein the adaptive smoothing means encodes and transmits a bit that indicates a type of filter characteristics for performing the smoothing. 
     
     
       13. The image coding apparatus as defined in  claim 1  wherein a bit that indicates a presence or absence of smoothing by the smoothing means is encoded and transmitted. 
     
     
       14. An image decoding apparatus comprising:
 motion compensation predicting means for generating a motion-compensated prediction image by detecting movement amounts in predetermined first partial image regions with respect to a reference image; 
 smoothing means for performing smoothing of pixels located at the boundaries of adjoining images of the first partial image regions, with respect to the motion-compensated prediction image obtained by the motion compensation predicting means; 
 prediction residual decoding means for decoding a prediction residual signal from an encoding apparatus; and 
 adding means for obtaining a decoded image by adding a decoded prediction residual signal obtained by the prediction residual decoding means, and a smoothed prediction image from the smoothing means, 
 wherein the smoothing means further comprises,
 activity level setting means for determining an activity level of second partial image regions; 
 adaptive smoothing means for establishing an intensity of the smoothing on a basis of the activity level set with said activity level setting means, and for performing smoothing of pixels located at the boundaries between adjoining images of the first partial image regions; and 
 post-processing means for performing suppression processing of a result of smoothing by using a threshold value established in accordance with said activity level, 
 
 wherein the first partial image regions are larger in size than the second partial image regions, so that at least one second partial image region fits into one of the first partial image regions, and the second partial image regions overlap a boundary of two adjacent first partial image regions when determining said activity level by the activity level setting means, and the second partial image regions are arranged in a cross-like pattern, having a central block with two blocks arranged above and below the central block, and two other blocks arranged to the left and the right of the central block. 
 
     
     
       15. The image decoding apparatus as defined in  claim 14  wherein the motion compensation predicting means obtains the prediction image by using decoded local decoding image data as the reference image. 
     
     
       16. The image decoding apparatus as defined in  claim 14  wherein the smoothing means performs the smoothing in accordance with a difference value for the movement amount between the adjoining images of the first partial image regions. 
     
     
       17. The image decoding apparatus as defined in  claim 14  wherein the smoothing means performs the smoothing in accordance with a difference in a decoding method between the adjoining images of the first partial image regions. 
     
     
       18. The image decoding apparatus as defined in  claim 14  wherein the smoothing means performs the smoothing based on whether the adjoining images of the first partial image regions have been predicted by bidirectional prediction, by unidirectional prediction, or by I-frame prediction. 
     
     
       19. The image decoding apparatus as defined in  claim 14  wherein the smoothing means performs the smoothing in accordance with an error margin amount between the prediction image prior to smoothing, and the reference image from which the prediction image obtained by the motion compensation predicting means is generated. 
     
     
       20. The image decoding apparatus as defined in  claim 14 , wherein said activity level of the second partial image regions is determined on a basis of at least one of: a difference value for the movement amount between the adjoining images of the first partial image regions, a difference in coding method between the adjoining images of the first partial image regions, a difference in image prediction method between the adjoining images of the first partial image regions, and an error margin amount between the prediction image prior to smoothing, and the reference image from which the prediction image obtained by the motion compensation predicting means is generated. 
     
     
       21. The image decoding apparatus as defined in  claim 20  wherein the activity level setting means sets the activity level for individual second partial image regions and inputs to the adaptive smoothing means the larger value of values for the activity level obtained by comparing the first partial image regions which are subjected to smoothing. 
     
     
       22. The image decoding apparatus as defined in  claim 20  wherein the activity level setting means sets the activity level for boundaries between the first partial image regions and inputs the activity level to the adaptive smoothing means. 
     
     
       23. The image decoding apparatus as defined in  claim 20  wherein the adaptive smoothing means changes the number of pixels subjected to smoothing in accordance with the activity level obtained by the activity level setting means. 
     
     
       24. The image decoding apparatus as defined in  claim 20  wherein the adaptive smoothing means switches filter characteristics for performing the smoothing in accordance with the activity level obtained by the activity level setting means. 
     
     
       25. The image decoding apparatus as defined in  claim 24  wherein the adaptive smoothing means switches the filter characteristics on a basis of a bit that indicates a type of filter characteristics for performing the smoothing, the bit being decoded from compressed input data from the encoding apparatus. 
     
     
       26. The image decoding apparatus as defined in  claim 14  wherein the smoothing processing is controlled on a basis of a bit that indicates a presence or absence of the smoothing, the bit being decoded from compressed input data. 
     
     
       27. An image coding method performed on a image coding apparatus, comprising:
 a motion compensation predicting step performed on a motion detection unit of the image coding apparatus of generating a motion-compensated prediction image by detecting movement amounts in predetermined first partial image regions of an input image with respect to a reference image stored in a frame memory; 
 a smoothing step performed on a filter unit of the image coding apparatus of performing smoothing of pixels located at the boundaries of adjoining images of the first partial image regions, with respect to the motion-compensated prediction image obtained by the motion compensation predicting step; and 
 a prediction residual coding step of coding a prediction residual signal obtained from the difference between the input image and the smoothed prediction image, 
 wherein the smoothing step further comprises,
 an activity level setting step of determining an activity level of second partial image regions; 
 an adaptive smoothing step of establishing an intensity of the smoothing on a basis of the activity level set with said activity level setting means, and of performing smoothing of pixels located at the boundaries between adjoining images of the first partial image regions; and 
 a post-processing step of performing suppression processing of a result of the smoothing by using a threshold value established in accordance with the activity level, 
 
 wherein the first partial image regions are larger in size than the second partial image regions, so that at least one second partial image region fits into one of the first partial image regions, and the second partial image regions overlap a boundary of two adjacent first partial image regions when determining said activity level by the activity level setting step, and the second partial image regions are arranged in a cross-like pattern, having a central block with two blocks arranged above and below the central block, and two other blocks arranged to the left and the right of the central block. 
 
     
     
       28. The image coding method performed on a image coding apparatus as defined in  claim 27  wherein the motion compensation predicting step detects the movement amount by using encoded local decoding image data as the reference image. 
     
     
       29. The image coding method performed on a image coding apparatus as defined in  claim 27  wherein the smoothing step performs the smoothing in accordance with a difference value for the movement amount between the adjoining images of the first partial image regions. 
     
     
       30. The image coding method performed on a image coding apparatus as defined in  claim 27 , wherein the smoothing step performs the smoothing based on whether the adjoining images of the first partial image regions have been predicted by bidirectional prediction, by unidirectional prediction, or by I-frame prediction. 
     
     
       31. The image coding method performed on a image coding apparatus as defined in  claim 27  wherein the smoothing step performs the smoothing in accordance with a difference in a image prediction method between the adjoining images of the first partial image regions. 
     
     
       32. The image coding method performed on a image coding apparatus as defined in  claim 27  wherein the smoothing step performs the smoothing in accordance with an error margin amount between the prediction image prior to smoothing, and the reference image from which the prediction image obtained by the motion compensation predicting step is generated. 
     
     
       33. The image coding method performed on a image coding apparatus as defined in  claim 27 , wherein said activity level of the second partial image regions is determined on a basis of at least one of: a difference value for the movement amount between the adjoining images of the first partial image regions, a difference in coding method between the adjoining images of the first partial image regions, a difference in image prediction method between the adjoining images of the first partial image regions, and an error margin amount between the prediction image prior to smoothing, and the reference image from which the prediction image obtained by the motion compensation predicting step is generated. 
     
     
       34. The image coding method performed on a image coding apparatus as defined in  claim 33  wherein the activity level setting step sets the activity level for individual second partial image regions and inputs to the adaptive smoothing step the larger value of values for the activity level obtained by comparing the first partial image regions which are subjected to smoothing. 
     
     
       35. The image coding method performed on a image coding apparatus as defined in  claim 33  wherein the activity level setting step sets the activity level for boundaries between the first partial image regions and inputs the activity level to the adaptive smoothing step. 
     
     
       36. The image coding method performed on a image coding apparatus as defined in  claim 33  wherein the adaptive smoothing step changes the number of pixels subjected to smoothing in accordance with the activity level obtained by the activity level setting step. 
     
     
       37. The image coding method performed on a image coding apparatus as defined in  claim 33  wherein the adaptive smoothing step switches filter characteristics for performing the smoothing in accordance with the activity level obtained by the activity level setting step. 
     
     
       38. The image coding method performed on a image coding apparatus as defined in  claim 37  wherein the adaptive smoothing step encodes and transmits a bit that indicates a type of filter characteristics for performing the smoothing. 
     
     
       39. The image coding method performed on a image coding apparatus as defined in  claim 27  wherein a bit that indicates a presence or absence of smoothing by the smoothing step is encoded and transmitted. 
     
     
       40. An image decoding method performed on a image coding apparatus comprising:
 a motion compensation predicting step performed on a motion detection unit of the image coding apparatus of generating a motion-compensated prediction image by detecting movement amounts in predetermined first partial image regions with respect to a reference image stored in a frame memory; 
 a smoothing step performed on a filter unit of the image coding apparatus of performing smoothing of pixels located at the boundaries of adjoining images of the first partial image regions, with respect to the motion-compensated prediction image obtained by the motion compensation predicting step; 
 a prediction residual decoding step of decoding a prediction residual signal from an encoding apparatus; and 
 an adding step of obtaining a decoded image by adding a decoded prediction residual signal obtained by the prediction residual decoding step, and the smoothed prediction image obtained by the smoothing step, 
 wherein the smoothing step further comprises,
 an activity level setting step of determining an activity level of second partial image regions; 
 an adaptive smoothing step of establishing an intensity of the smoothing on a basis of the activity level set with said activity level setting step, and of performing smoothing of pixels located at the boundaries between adjoining images of the first partial image regions; and 
 a post-processing step of performing suppression processing of a results of smoothing by using a threshold value established in accordance with said activity level, 
 
 wherein the first partial image regions are larger in size than the second partial image regions, so that at least one second partial image region fits into one of the first partial image regions, and the second partial image regions overlap a boundary of two adjacent first partial image regions when determining said activity level by the activity level setting step, and the second partial image regions are arranged in a cross-like pattern, having a central block with two blocks arranged above and below the central block, and two other blocks arranged to the left and the right of the central block. 
 
     
     
       41. The image decoding method performed on a image coding apparatus as defined in  claim 40  wherein the motion compensation predicting step obtains the prediction image by using decoded local decoding image data as the reference image. 
     
     
       42. The image decoding method performed on a image coding apparatus as defined in  claim 40  wherein the smoothing step performs the smoothing in accordance with a difference value for the movement amount between the adjoining images of the first partial image regions. 
     
     
       43. The image decoding method performed on a image coding apparatus as defined in  claim 40  wherein the smoothing step performs the smoothing based on whether the adjoining images of the first partial image regions have been predicted by bidirectional prediction, by unidirectional prediction, or by I-frame prediction. 
     
     
       44. The image decoding method performed on a image coding apparatus as defined in  claim 40  wherein the smoothing step performs the smoothing in accordance with a difference in image prediction method between the adjoining images of the first partial image regions. 
     
     
       45. The image decoding method performed on a image coding apparatus as defined in  claim 40  wherein the smoothing step performs the smoothing in accordance with an error margin amount between the prediction image prior to smoothing, and the reference image from which the prediction image obtained by the motion compensation predicting step is generated. 
     
     
       46. The image decoding method performed on a image coding apparatus as defined in  claim 40 , wherein said activity level of the second partial image regions is determined on a basis of at least one of: a difference value for the movement amount between the adjoining images of the first partial image regions, a difference in coding method between the adjoining images of the first partial image regions, a difference in image prediction method between the adjoining images of the first partial image regions, and an error margin amount between the prediction image prior to smoothing, and the reference image from which the prediction image obtained by the motion compensation predicting step is generated. 
     
     
       47. The image decoding method performed on a image coding apparatus as defined in  claim 46  wherein the activity level setting step sets the activity level for individual second partial image regions and inputs to the adaptive smoothing step the larger value of values for the activity level obtained by comparing the first partial image regions which are subjected to smoothing. 
     
     
       48. The image decoding method performed on a image coding apparatus as defined in  claim 46  wherein the activity level setting step sets the activity level for boundaries between the first partial image regions and inputs the activity level to the adaptive smoothing step. 
     
     
       49. The image decoding method performed on a image coding apparatus as defined in  claim 46  wherein the adaptive smoothing step changes the number of pixels subjected to smoothing in accordance with the activity level obtained by the activity level setting step. 
     
     
       50. The image decoding method performed on a image coding apparatus as defined in  claim 46  wherein the adaptive smoothing step switches filter characteristics for performing the smoothing in accordance with the activity level obtained by the activity level setting step. 
     
     
       51. The image decoding method performed on a image coding apparatus as defined in  claim 50  wherein the adaptive smoothing step switches the filter characteristics on a basis of a bit that indicates a type of filter characteristics for performing the smoothing, the bit being decoded from compressed input data from the encoding apparatus. 
     
     
       52. The image decoding method performed on a image coding apparatus as defined in  claim 40  wherein the smoothing processing is controlled on a basis of a bit that indicates a presence or absence of the smoothing, the bit being decoded from the compressed input data. 
     
     
       53. The image coding apparatus as defined in  claim 1 , wherein the activity level setting means determines a direction of a prediction for the blocks above, below, left, right of the central block, and determines a direction of prediction for the central block, and determines if the direction of prediction is the same for all the blocks. 
     
     
       54. The image decoding apparatus as defined in  claim 14 , wherein the activity level setting means determines a direction of a prediction for the blocks above, below, left, right of the central block, and determines a direction of prediction for the central block, and determines if the direction of prediction is the same for all the blocks. 
     
     
       55. The image coding method performed on a image coding apparatus as defined in  claim 27 , wherein the activity level setting step determines a direction of a prediction for the blocks above, below, left, right of the central block, and determines a direction of prediction for the central block, and determines if the direction of prediction is the same for all the blocks. 
     
     
       56. The image decoding method performed on a image coding apparatus as defined in  claim 40 , wherein the activity level setting step determines a direction of a prediction for the blocks above, below, left, right of the central block, and determines a direction of prediction for the central block, and determines if the direction of prediction is the same for all the blocks.